Using data analytics and laboratory experiments to advance the understanding of reservoir rock properties


TR Number



Journal Title

Journal ISSN

Volume Title


Virginia Tech


Conventional and unconventional reservoirs are both critical in oilfield developments. After waterflooding treatments over decades, the petrophysical properties of a conventional reservoir may change in many aspects. It is crucial to identify the variations of these petrophysical properties after the long-term waterflooding treatments, both at the pore and core scales. For unconventional reservoirs, the productivity and performance of hydraulic fracturing in shales are challenging because of the complicated petrophysical properties. The confining pressure imposed on a shale formation has a tremendous impact on the permeability of the rock. The correlation between confining pressure and rock permeability is complicated and might be nonlinear. In this thesis, a series of laboratory tests was conducted on core samples extracted from four U.S. shale formations to measure their petrophysical properties. In addition, a special 2D microfluidic equipment that simulates the pore structure of a sandstone formation was developed to investigate the influence of injection flow rate on the development of high-permeability flow channels. Moreover, the multiple linear regression (MLR) model was applied with the predictors based on the development stages to quantify the variations of reservoir petrophysical properties. The MLR model outcome indicated that certain variables were effectively correlated to the permeability. The 2D microfluidic model demonstrated the development of viscous fingering when the injection water flow rate was higher than a certain level, which resulted in reduced overall sweep efficiency. These comprehensive laboratory experiments demonstrate the role of confining pressure, Klinkenberg effect, and bedding plane direction on the gas flow in the nanoscale pore space in shales.



rock properties, in-lab experiment, Klinkenberg effect, multiple linear regression